0

Dynamic Product Recommendation System

Developed a personalized dynamic recommendation system using NLP, Product2Vec, and other machine learning models to improve product recommendations at Amazon.

Website

Dynamic Product Recommendation System at Amazon

During my Software Developer Engineering Internship at Amazon, I worked on transforming the static recommendation system into a personalized dynamic recommendation system.

Key Achievements

Advanced Machine Learning Models

Utilized Natural Language Processing (NLP), Product2Vec, and other machine learning models to create personalized product recommendations, moving away from a static recommendation system to a more dynamic and personalized approach.

Scalable Graph Database

Created a low latency, scalable graph database using AWS Neptune with 100 million nodes and 5 billion edges to retrieve data at run time. This implementation achieved a 25% faster retrieval time than the previous implementation, resulting in a 5% increase in conversion rate.

AWS Integration

Leveraged various AWS services, including Neptune, S3, SageMaker, and DRS, to construct the graph database. Developed comprehensive tests to benchmark the database performance using EC2 instances and the open-source software Locust.

Technologies Used

  • AWS Neptune (Graph Database)
  • AWS S3, SageMaker, DRS
  • Natural Language Processing (NLP)
  • Product2Vec
  • Machine Learning
  • EC2
  • Locust (Performance Testing)